Antibiotic adverse events on an outpatient parenteral antibiotic service: a retrospective cohort study

Abstract

Objectives

To identify the rate of adverse events for ceftriaxone, ertapenem, daptomycin, teicoplanin, piperacillin/tazobactam and meropenem on an outpatient parenteral antibiotic (OPAT) service.

Method

We identified all patients on our prospectively populated OPAT Access Database from 2009 to 2016. Duration on OPAT and the details of any documented antibiotic adverse/toxic events, were recorded.

Results

Ertapenem was used most commonly at 356/961 (37%) specific antibiotic episode per all antibiotic episodes, followed jointly by both ceftriaxone 241/961 (25%), teicoplanin 237/961 (25%), with meropenem 59/961 (6%), piperacillin/tazobactam 42/961 (4%) and daptomycin 26/961 (3%). The total number of antibiotic adverse events for the six antibiotics being investigated was 48/961 (5%). Piperacillin/tazobactam has a significantly higher rate of adverse events compared with most of the other antibiotics studied (12/42; 29%) followed by daptomycin (3/26; 12%; p=0.1362), ceftriaxone (16/241; 7%; p=0.0001), meropenem (3/59; 5%; p=0.0015), teicoplanin (10/237; 4%; p=0.0001) and ertapenem (4/356; 1%; p=0.0001). Ertapenem has a significantly lower rate of toxicity compared with some of the other antibiotics investigated. When standardised per 1000 OPAT days, piperacillin/tazobactam continues to have a significantly higher rate of toxicity (8.3 episodes of antibiotic toxicity per 1000 OPAT days), with again ertapenem having a significant lower rate of toxicity (0.5) compared with some of the remaining antibiotics.

Conclusion

While the overall number of adverse events directly attributable to the antibiotics examined was only 5%, it may be possible to optimise antibiotics to reduce this percentage.

Introduction

The outpatient parenteral antimicrobial service (OPAT) was established as a means to administer intravenous antibiotics in the community for specified cases where oral antibiotics were not satisfactory. The obvious benefits are: increased capacity for treatment; increased availability of hospital beds; reduced hospital days; promoting the well-being of patients by allowing treatment in the comfort of their own home and limiting the negative psycho-social impact of long hospitalisation.

As the service successes have increased over the decade, it has expanded from the treatment of simpler conditions such as cellulitis to providing treatment for more complex conditions such as metalwork-associated osteomyelitis and vascular graft infections. Recent successes have seen multidrug-resistant and extensively drug-resistant patients with tuberculosis managed on the service in addition to a range of fungal conditions.

In order to administer antimicrobials, long lines are required, which allow venous access over longer durations than a simple cannula. These have their own well-recognised complications which include line infection and thrombosis.^1–3 These complications can be mitigated by regular examination of lines and sites^{3 4} and by administration of prophylactic low molecular weight heparin, although the effectiveness of the later is unclear and requires further study.^{5 6}

Of the factors which are determined to establish patient suitability for OPAT is whether their in-hospital antibiotic regime can be switched to an effective once-daily regime. Although not impossible to arrange twice-daily or three times daily administration, the logistics are understandably more difficult. However, there are circumstances where there are no other options, such as in the treatment of Pseudomonas-related infections with either piperacillin/tazobactam or meropenem, both three times daily regimens. For the once-daily regimens, the common antibiotic options include: ceftriaxone, ertapenem, teicoplanin and daptomycin. As such, these antibiotics are most commonly used in the OPAT service.

While studies have been carried out using these antibiotics looking at adverse effects,^7–9 few studies^{10 11} have looked into the long duration toxic effects, as used on OPAT.

In this study, we aimed to identify the rate of toxicity/adverse events of each of the the most common antibiotics used in our OPAT service: ceftriaxone, ertapenem, daptomycin, teicoplanin, piperacillin/tazobactam and meropenem. We wished to address the question of whether antibiotic toxicity should be taken into account when constructing an antibiotic regimen for OPAT patients.

Patients and method

In this single-centre retrospective cohort study, we identified all patients on our prospectively populated OPAT Access Database from 2009 to 2016. Those patients who had a documented antibiotic adverse/toxic event, the nature of the adverse event where specified, and duration on OPAT was recorded. In addition, individual patient demographics and initial diagnosis/condition were recorded.

Definitions

The definition of an antibiotic adverse/toxic event for the purpose of this study was taken from the WHO definition of an adverse reaction to a drug: a reaction that is ‘noxious, is unintended, and occurs at doses normally used in man’.¹²

Abnormal liver biochemistry was defined as those values above or below the limits of normal function attributed to the antibiotic.

Abnormal urea and electrolytes biochemistry was defined as electrolytes (potassium, sodium, corrected calcium, phosphate, magnesium) above and below the limits of normal function attributed to the antibiotic. For urea and creatinine, abnormal biochemistry was defined as urea and creatinine above the upper limit of normal for the individual patient or for those with chronic kidney disease a ‘50% or greater rise in serum creatinine’.¹³

Neutropenia defined as a neutrophil count below the lower limit of normal attributed to the antibiotic.

An unrecorded reaction was defined as a reaction attributed to the antibiotic in use, which was unspecified.

Statistical analysis

Rate of antibiotic toxicity has been expressed as a percentage and as a rate per 1000 OPAT days. Using the latter allows standardisation per 1000 OPAT days accounting for the variable duration of antibiotic episodes.

The χ² test and Fisher’s exact test were used to determine statistical significance when comparing antibiotic groups and toxicity. An unpaired t-test was used to determine the statistical significance between the ages of the two groups. Statistical analysis was performed using Graphpad QuickCalcs. A p value ≤0.05 was considered significant for calculation of significance between the ages of the groups with or without antibiotic adverse events. A p value <0.003 was considered significant for the rates (percentage and per 1000 OPAT days) between antibiotic adverse events. This threshold was calculated using a Bonferroni correction with initial p value <0.05 and 15 tests of significance performed.

Results

Antibiotics

There were 1269 treatment episodes recorded on our OPAT database, of which 961 episodes involved the use of at least one of the six antibiotics being investigated (ceftriaxone, ertapenem, daptomycin, teicoplanin, piperacillin/tazobactam and meropenem). Ertapenem was most commonly used, in 37% of episodes; ceftriaxone and teicoplanin were also commonly used, each accounting for 25% of episodes; meropenem (6%), piperacillin/tazobactam (4%) and daptomycin (3%) accounted for far fewer episodes.

Demographics/initial diagnoses

In the wider group, those without an antibiotic adverse event, the average age was 60 years, with the median in 60–69 age group. The male-to-female ratio was 2:1. Of those who had an antibiotic adverse event the average age was 56 years, with the median in 50–59 age group. There was no statistical significance in ages between the two groups (p=0.1310). The male-to-female ratio was 1.4:1. The most common initial diagnoses in the antibiotic toxicity group included: osteomyelitis (14/48), discitis (3/48), septic arthritis, prosthetic joint infections (4/48), malignant otitis media (4/48), diabetic foot infections (2/48), vascular graft infections (2/48) and endocarditis (3/48).

Duration on OPAT

The total duration on OPAT for the six antibiotics was 25 075 OPAT days. The OPAT days per antibiotic were ceftriaxone 5500 days (22%), daptomycin 739 (3%), ertapenem 8409 (34%), meropenem 2239 (9%), piperacillin/tazobactam 1453 (6%) and teicoplanin 6735 days (27%). As expected, the number of OPAT days per antibiotic broadly correlated with the relative proportion of antibiotic episodes.

The mean and median duration on OPAT respectively for each antibiotic was: ceftriaxone, 24 and 22 days; daptomycin, 28 and 17.5; ertapenem 26 and 21; meropenem, 39 and 28; piperacillin/tazobactam 35 and 27.5; and teicoplanin, 33 and 33 days.

Antibiotic adverse events

The total number of antibiotic adverse events for the six antibiotics being investigated was 48/961 (5%).

Piperacillin/tazobactam had a significantly higher rate of adverse events than most of the other antibiotics studied compared with daptomycin (p=0.1362), ceftriaxone (p=0.0001), meropenem (p=0.0015), teicoplanin (p=0.0001) and ertapenem (p=0.0001) (table 1).

Table 1.

Number and percentage adverse events (AEs) with corresponding p values using the χ² test and Fisher’s exact test

	Number of AEs	% AEs	P values
			Ceft	Dapt	Erta	Mero	Pip/Taz	Teic
Ceft	16	7		0.4105	0.0003	1.0000	0.0001	0.3138
Dapt	3	12			0.0082	0.3642	0.1362	0.1257
Erta	4	1				0.0628	0.0001	0.0239
Mero	3	5					0.0015	0.7274
Pip/Taz	12	29						0.0001
Teic	10	4
Total	48

A p value <0.003 was considered to be statistically significant.

Ceft, ceftriaxone; Dapt, daptomycin; Erta, ertapenem; Mero, meropenem; Pip/Taz, piperacillin/tazobactam; Teic, teicoplanin.

Ertapenem has a significantly lower rate of toxicity compared with some of the other antibiotics: daptomycin (p=0.0082); ceftriaxone (p=0.0003); meropenem (p=0.0628); piperacillin/tazobactam (p=0.0001) and teicoplanin (p=0.0239) (table 1).

Differences in antibiotic adverse events between the remaining antibiotics (ceftriaxone, meropenem, daptomycin, teicoplanin) were not statistically significant.

By standardising the results per 1000 OPAT days, we find a similar picture: piperacillin/tazobactam has a statistically significantly higher rate of toxicity (8.3 antibiotic toxicity per 1000 OPAT days). Again, ertapenem has a statistically significant lower rate of toxicity (0.5) compared with some of the remaining antibiotics. The only differences were that ceftriaxone (2.9) now had a higher rate of toxicity than daptomycin (2.7), although again not statistically significant (table 2).

Table 2.

Comparison of antibiotic adverse events (AEs) per 1000 OPAT days with corresponding p values using the χ² test and Fisher’s exact test

	Antibiotic AEs per 1000 OPAT days	P values
		Ceft	Dapt	Erta	Mero	Pip/Taz	Teic
Ceft	2.9		0.7809	0.0001	0.3308	0.0001	0.1483
Dapt	2.7			0.0001	0.2465	0.0001	0.074
Erta	0.5				0.0069	0.0001	0.0276
Mero	1.3					0.0001	0.692
Pip/Taz	8.3						0.0001
Teic	1.3

A p value <0.003 was considered to be statistically significant.

Ceft, ceftriaxone; Dapt, daptomycin; Erta, ertapenem; Mero, meropenem; OPAT, outpatient parenteral antibiotic; Pip/Taz, piperacillin/tazobactam; Teic, teicoplanin.

From our data, we can split these six antibiotics into three groups. Group 1 contains piperacillin/tazobactam with higher rate of adverse events (29%). Group 2 contains daptomycin, ceftriaxone, meropenem and teicoplanin with the middle range of antibiotic adverse events (4%–12%). Finally, group 3 contains ertapenem which appeared the safest (1%).

Type of adverse event

The specific adverse events observed were rashes, liver function derangement, electrolyte, urea and creatinine disturbance, neutropenia and unrecorded. These were therefore the categories recorded in addition to other rarer reactions (see table 3 for type of adverse event per antibiotic).

Table 3.

Number of adverse events (AEs) per type of reaction

	AE episodes	Type of adverse event
		Rash	Abnormal LFT	Abnormal U&E	Neutropenic	Unrecorded reaction	Other
Ceft	16	4	2	1	5	4	0
Dapt	3	1	0	1	1	0	0
Erta	4	0	0	0	1	2	1
Mero	3	0	1	0	1	1	0
Pip/Taz	12	5	0	0	0	5	2
Teic	10	4	0	0	0	4	2
Total	48	14	3	2	8	16	5

Ceft, ceftriaxone; Dapt, daptomycin; Erta, ertapenem; LFT, liver function test; Mero, meropenem; Pip/Taz, piperacillin/tazobactam; Teic, teicoplanin; U&E, urea and electrolytes.

Discussion

The total number of antibiotic adverse events in this study was 48/961 (5%), which compared with the 25%¹⁴ quoted in the literature is relatively small. Although in the later study a significantly wider range of antimicrobials were tested including antivirals and antifungals. In addition, it is likely that the value in this study is an underestimate given the observation bias of antibiotic reactions necessitating a switch in therapy.

Piperacillin/tazobactam versus meropenem

It is clear from our study that piperacillin/tazobactam had a significantly higher rate of adverse events than most of the other antibiotics investigated. Piperacillin/tazobactam was rarely chosen (4%) as a parenteral agent on OPAT. The reason for this is the three times daily dosing regimen required which has obvious logistical problems in the community. The usual indication is in the treatment of confirmed Pseudomonas infections, or conditions such as malignant otitis externa and bronchiectasis, where empirical treatment should include Pseudomonas cover. Other intravenous antibiotics with Pseudomonas cover which have been used on OPAT include ceftazidime and meropenem. Ceftazidime was not investigated in this particular study due to its limited use. However, meropenem was found to have significantly fewer adverse events when standardised per 1000 OPAT days. Given this is the case, whether meropenem should be used first line rather than piperacillin/tazobactam in certain specific circumstances outlined should be considered. This would need to be carefully balanced as the reasons against this are compelling, namely antimicrobial stewardship: the importance of limiting carbapenem use in order to prevent the selection of resistance to an important antibiotic of last resort.¹⁵ The benefit of limiting meropenem is both to the individual patient, preventing resistant organisms being selected making further treatment courses difficult, especially in cases such malignant otitis media and basal skull osteomyelitis where prolonged courses are sometimes required¹⁶; and to the general public in limiting the spread of resistant organisms. It is possible that dosage frequency may have contributed to an increase in rate of adverse events, including events that may be related to infusion rates. This may be disproportionally higher in high frequency dosing regimens including piperacillin/tazobactam with a three times daily regimen in comparison to ertapenem, ceftriaxone, teicoplanin, daptomycin with once-daily regimens. However, meropenem which has a similar regimen to piperacillin/tazobactam had a significantly lower rate of adverse events in comparison.

Furthermore, whether there are cases where an oral antibiotic such as ciprofloxacin might be more appropriate needs to be considered, given its relatively good bone penetration and decades old studies suggesting efficacy.^17–19

Ertapenem

Ertapenem was the most common antibiotic administered on OPAT. This is probably a result of its wide spectrum of activity covering both Gram-positive and Gram-negative aerobes and anaerobes, and its perceived good safety profile,^{10 20} confirmed by this study. It is unclear why ertapenem appears to have a lower toxicity profile than other antibiotics. Discriminating factors and thus possible explanations may include: a lower effective dose and reduced frequency of administration than most other intravenous antibiotics; and structural differences in comparison to other beta-lactam antibiotics.²¹ In terms of alternative once-daily regimens, a similar spectrum of activity is found with ceftriaxone. Regarding ertapenem versus ceftriaxone use in Gram-negative infections, while the efficacy appears to be similar,²² it is likely the emergence of resistance and specifically beta-lactamases have made ertapenem a more frequent choice,²³ in addition to the perceived risk of Clostridium difficile infections associated with cephalosporins.²⁴ However, of note there were no documented instance of C. difficile infection despite 241 instances of ceftriaxone use, and a total 5500 days on OPAT. Given the high use of ertapenem on our service and the importance of preventing selection of carbapenem resistance, a case could be made to increase the use of ceftriaxone. However, as with meropenem above, the carbapenem may still be preferred as there were significantly fewer associated adverse events.

Ceftriaxone, teicoplanin, daptomycin and meropenem

While there were observable differences in rate of toxicity between these antibiotics, these differences were not significant by statistical analysis. Moreover, these antibiotics seemed to form a middle group of antibiotic toxicity rates between piperacillin/tazobactam and ertapenem.

Type of adverse event

While there was a general mix of adverse reactions, given the low numbers in each specific category, it is difficult to draw statistically significant conclusions when comparing antibiotics—but overall a rash was the the most commonly described reaction. Rash was most commonly associated with piperacillin/tazobactam and teicoplanin. Neutropenia tended to be caused by ceftriaxone, daptomycin, teicoplanin and piperacillin/tazobactam. A better understanding of the relative rates of adverse events is particularly important in clinical practice when multiple antibiotics are in use or in close proximity of use, in order to determine the likely causative agent; preventing a less desirable antibiotic regimen where multiple antibiotic classes have been summarily excluded.

Overall, there were a large number of reactions of which the specifics were not recorded. Given this was a retrospective study, we were unable to investigate this further. This is certainly one of the limiting factors of this study design. This lack of specific data may have had different causes, including: simply the failure to record the nature of the reaction and doubt as to the aetiology of reactions; factors which may have contributed to an overestimation of adverse events.

By correlating adverse events per antibiotic with the literature, our study found a higher rate of adverse events with ceftriaxone (5.8%^{9 25}), meropenem (3%⁷), piperacillin/tazobactam (4%²⁶), daptomycin (3%⁹) and teicoplanin (10%⁸). This could be as a result of the extended durations of intravenous antibiotics on an OPAT regimen (as in our study) compared with an in-hospital regimen.

Ertapenem was the only antibiotic to have a lower rate of adverse events (1.2%⁹). In addition, in regard to aetiology of adverse events, we found no seizure-related adverse events previously associated with meropenem or ertapenem⁹. The rate of neutropenia associated with ceftriaxone was higher than the 0.3%⁹ quoted in the literature. Again, this may have been related to the probable extended durations of intravenous antibiotics on OPAT compared with an in-hospital regimen.

This was a retrospective study with all the usual difficulties associated with such a study design. However, there are firm conclusions which can be made. The study inclusion criteria were broad and did not differentiate between the condition, severity or organism being treated. Nevertheless, it is not clear how this would have changed the rate of toxicity, except for the possibility of adverse events which may have been associated with the condition or organism. The second problem, inherent in retrospective studies, is the reliance on data from the observer and clinical accuracy at the time of the event. This was evident from the documented number of unrecorded reactions. In addition, the rate of adverse events for each antibiotic is likely an underestimate. For example, while ertapenem maintained a low rate of adverse events, these adverse events (1%) were lower than that quoted in the literature (3%).¹⁰ While every effort has been made to record adverse events on the OPAT database in real-time, there will be inevitable events which have not been recorded. These events are most likely mild in origin, have not affected the antibiotic administration and have been tolerated by the patient. These adverse events may include managed diarrhoea, explaining the reduced frequency of gastrointestinal side effects.^{10 20} However, it is expected that events requiring antibiotic change have been recorded accurately.

The conditions being treated on OPAT were wide and varied. As a necessity, these conditions must be stable in order to be managed in the community. Decisions on antibiotics were conducted according to best available practice. The decision-making process would typically include where possible the condition, organism and antibiotic sensitivity profile. Conditions where antibiotic choice and duration were not defined, were discussed in a multidisciplinary meeting. As such, certain conditions were more likely to receive certain antibiotics. In addition, certain conditions were treated for a longer duration than others. It is possible that the conditions themselves may have contributed to the number of adverse events. Causality was clinically judged in real-time, therefore due to retrospective study design, it was not possible to retrospectively confirm causality. Thus, this study was reliant on initial clinical judgement and data entry. It is possible that these factors may have contributed to the number of adverse effects. In addition, time to adverse event was not collected at the time of data entry, this may have provided greater insight as to whether duration of antibiotics may have been a key factor.

Conclusion

While the overall number of adverse events directly attributable to the antibiotics examined was only 5%, it is possible to optimise antibiotics further to reduce this percentage. However, further optimisation may be at the expense of appropriate antimicrobial stewardship and ultimately may not be in the best interest of the patient. Our data suggest that there is a significantly higher rate of toxicity with piperacillin/tazobactam, compared with ceftriaxone, teicoplanin, meropenem and ertapenem. As a result, alternative antibiotics may be preferable. In addition, we noted a significantly lower rate of toxicity for ertapenem in comparison to piperacillin/tazobactam and ceftriaxone. As a result, ertapenem may be the ideal antibiotic to reduce adverse effects for multiple conditions with its broad spectrum of activity.

Ultimately, antibiotic adverse events, especially with the case of piperacillin/tazobactam may play a role in choice of antibiotics in a given situation. This must be judged with the necessity of antibiotics resistance patterns, local drug penetrance and prevention of acquiring resistance.

What this paper adds.

What is already known on this subject

• Ertapenem has a low rate of adverse events.

• Minimal studies looking at extended duration of antibiotics as used on outpatient parenteral antibiotic.

What this study adds

• Evidence that ertapenem over an extended duration has a lower rate of adverse events over other antibiotics.

• Evidence that piperacillin/tazobactam over an extended duration has a higher rate of adverse events over other antibiotics.

Data availability statement

All data relevant to the study are included in the article.